Resident Communication With Patients and Families Preferring Languages Other Than English

This cross-sectional study took place between September 2019 and January 2020 at an urban, high-volume quaternary care pediatric hospital in the northeastern United States, which houses a large pediatric residency program. The project team consisted of resident physicians, attending physicians, inpatient care team assistants, language services leadership, and senior research advisors. Inpatient care team assistants are individuals who are assigned to inpatient teams to assist with various administrative tasks, such as scheduling follow-up appointments and obtaining patient records; they can also facilitate scheduling interpreters and routinely assist in research activities. The 3 most frequently preferred languages of hospitalized patients and families at this institution are English, Spanish, and Arabic. This study was approved by the local institutional review board.

At the time of this study, 2 modes of interpretation were readily available to inpatient clinicians: telephonic interpreters, in virtually all languages, and in-person interpreters, primarily in Spanish and Arabic. In-person interpreters in other languages could also be facilitated with advance planning. Video interpretation was not readily available in inpatient units.

All clinicians had access to telephonic interpretation at all times via hospital-issued mobile phones as well as landline phones available in all patient rooms. In-person interpreters could be scheduled via two primary modalities: placing an order in the electronic medical record or calling interpreter services directly. Anyone could request an interpreter, including nurses, inpatient care team assistants, residents, and families. At the time of this study, there was no universal, standardized process for scheduling interpreters for hospitalized patients and families.

To address our study’s first aim, characterizing residents’ perceived behaviors, comfort, and barriers regarding interpreter services and interactions with patients and families preferring LOE, we developed a 20-question anonymous electronic questionnaire (Supplemental Fig 2). Survey content was developed based on comprehensive literature review and input from local experts, including leaders in language services and educational research.^7–14  Local experts included leaders in language services and educational research. Participants were asked for demographic information as well as previous training or education regarding interpreter services. Subsequent sections contained series of Likert scales regarding resident comfort with and barriers to interpreter use, as well as resident behaviors related to interpretation. The survey was hosted in REDCap, a secure online application. We invited all pediatric residents to participate in the study via an e-mail that included the link to the survey. The survey was open for approximately 3 weeks, in October 2019, and 2 additional reminder e-mails were sent.

Our study’s second aim was to evaluate the impact of caregiver preferred language on frequency of resident communication events with patients and families. In collaboration with local experts, interpretation services, and resident stakeholders, a resident-patient communication tracking tool was developed, with the goal of quantifying resident communication events with patients and families (Fig 1). This tool asked residents to indicate the preferred language of their patients’ caregivers, whether the family was present on inpatient morning rounds, if the patient and family were updated after rounds by the resident, and, if so, how many times, and whether an interpreter was used for the update(s). Preferred language referenced the language preferred for medical updates by the patient’s primary caregiver. The decision was made to ask residents specifically about updates provided after rounds because resident stakeholders indicated that this was a time frame when communication with families is important, though at times can be difficult to prioritize with competing tasks. Residents indicated that it was an expectation that they would communicate at least once after morning rounds with their patients and their families.

The communication tool was administered to 6 inpatient resident teams. All teams were located on medical floors and consisted of a combination of general pediatrics and subspecialty patients. The subspecialties represented on these teams included endocrinology, complex care pediatrics, hematology, metabolism, neurology, pulmonology, and rheumatology. Residents on each team were either interns (postgraduate year 1 [PGY-1]) paired with seniors (PGY-2 and PGY-3), or solely senior residents (PGY-2). Each of these teams engages in a rounding model intended to be family-centered, in which the clinical team goes to a patient’s room to discuss the plan of care for the day, and caregivers and patients are included in the discussion, if possible. Additionally, residents are expected to be the primary communicators with patients and families on these teams by providing regular updates and modifying the plan of care as needed. We specifically selected these teams because of similarities in rounding models, nurse staffing, resident-patient care ratios, and patient acuity.

The communication tool was administered biweekly between October 2019 and January 2020. We randomly selected the day of the week to administer the tool to a given team, in an effort to minimize the Hawthorne effect, or changes in behaviors secondary to knowledge that a study was occurring.²⁴  Study team members administered the tool in-person between 5:00 pm and 6:00 pm on selected days, which is just before resident evening sign-out. We specifically only targeted “front-line clinician” residents to complete the tool, a term indicating the resident with primary clinical responsibility for a patient.

The process for tool administration was developed and refined by the study team and relevant stakeholders. A study team member or team assistant first began by individually asking all front-line clinician residents on a given team if they had any patients preferring LOE. If a resident answered yes, he or she was asked to fill out the communication tool form anonymously and privately, listing all their patients for that day, both those preferring LOE and those preferring English. This strategy was implemented to maximally capture interactions with patients and families preferring LOE. Residents were provided with verbal assurance that all responses were confidential. Resident participation was optional and voluntary.

We generated descriptive summary statistics, including frequencies and percentages from questionnaire results and resident-patient communication tracking tool data, using SPSS software (IBM SPSS Statistics, IBM Corporation, Chicago, IL).

To examine differences in presence on family-centered rounds and likelihood of an update after rounds, we developed logistic regression models, controlling for inpatient team and resident year of training. These variables were selected by the study team as important clinical factors. A Poisson regression model was used to evaluate for differences in number of updates after rounds. These analyses were completed using the software R, version 4.0.5.

The Strengthening the Reporting of Observational Studies in Epidemiology guidelines were used in preparation of the study manuscript.²⁵ 

Response rate to the survey regarding interpretation comfort, practices, and barriers was 47%, with 63 total responses. The majority (76%) reported having some previous training regarding interpreter use. Subsequent questions asked about resident comfort with the 2 readily available forms of interpretation: telephone and in-person. More than 80% of residents reported feeling either “very comfortable” or “somewhat comfortable” with telephone interpretation (82%); more than 90% reported similar feelings of comfort with in-person interpretation (91%) (Table 1).

Residents were also queried regarding perceived barriers to using interpretation. All residents (100%) reported that time constraints during the prerounding period were a barrier. Other frequently recognized barriers included time constraints during rounds (84%), unpredictable rounds schedules (83%), and availability of interpreter services (75%) (Table 1).

An additional section of the questionnaire focused on behaviors related to interpretation. A minority of residents, 23%, reported always or usually using interpretation for prerounds. A smaller percentage of residents, 9%, reported always or usually using interpretation during rounds. Regarding the after rounds period, 82% of residents noted always or usually using interpretation during this time. Finally, when asked about use of interpretation in general, 70% of residents reported rarely or never communicating with families preferring LOE without interpretation (Table 2).

The resident-patient communication tracking tool was completed by 36 unique residents, representing 135 unique patients and 151 total patient encounters, with a 95% completion rate. Among these encounters, 110 caregivers (73%) preferred English, whereas 41 (27%) preferred other languages. Spanish (13%) and Arabic (7%) were the most frequently represented non-English languages (Table 3).

Logistic regression demonstrated that patients and families preferring LOE were significantly less likely to be present on inpatient rounds compared with patients and families preferring English (odds ratio, 0.19; 95% confidence interval [CI], 0.07-0.40). After adjusting for inpatient team and resident year of training, this difference remained (adjusted odds ratio, 0.17; 95% CI, 0.07-0.39). Similarly, patients and families preferring LOE were significantly less likely to receive a resident update after morning rounds, compared with their counterparts preferring English (odds ratio, 0.28; 95% CI, 0.13-0.59; adjusted odds ratio 0.29; 95% CI, 0.13-0.62). We also performed a Poisson regression to evaluate for differences in the number of updates after rounds. This demonstrated that patients and families preferring LOE receive updates at a rate 55% lower than those preferring English (incidence rate ratio, 0.45; 95% CI, 0.30-0.69) (Table 4).

For the updates that did occur after rounds with families preferring LOE (n = 18), an appropriate form of interpretation was used 53% of the time. In the remaining instances, providers reported either not using an interpreter, acting as an interpreter themselves (though not certified), or using a family member to provide interpretation.

Our study illustrates that patients and families preferring languages other than English experience inequitable communication with resident physicians while hospitalized. When communication does occur with families preferring LOE, residents self-report underuse of appropriate interpretation modalities, most prominently during the prerounding and rounding portions of the day. Though residents reported frequently using appropriate interpretation during the “after rounds” period, results from the communication tracking tool, in real time, were not congruent, and demonstrated that appropriate interpreter modalities were only used about half of the time. Importantly, this study also demonstrates that fewer communication events occur for hospitalized patients and families preferring LOE, both in terms of presence on rounds and updates by residents after rounds. These results reveal both subjective and objective inequitable communication for patients and families preferring LOE and contribute to a growing body of research that seeks to describe and address inequities in care and communication for this population.

A major impetus for embarking on this study was the concept of patient “touchpoints.” Currently, in the literature, touchpoints are described as discrete patient encounters with the health care system, including visits to the emergency department, clinic visits, or inpatient stays.^26,27  We posit that there are also various touchpoints during an inpatient hospitalization that are key to optimizing patient communication and care. We define these as any contact a patient has with a health care provider during their time in the hospital; this could include a nurse obtaining vital signs, a respiratory therapist administering a nebulized medication, or, in the case of our study, a resident providing an update. All these interactions represent critical inpatient touchpoints for patients, ones that may be experienced differently, or less frequently, by patients who prefer languages other than English.

Measuring inpatient touchpoints is difficult for various reasons, the most obvious being that each of these interactions is not routinely documented in the electronic medical record. For patients preferring LOE, interpretation use could be used as a proxy; however, interpretation is underused in clinical care, making it an unreliable metric.^8–17  Moreover, use of interpretation data would also not allow for comparative analyses with patients who prefer English. Videorecording health care encounters has shown promise, though this strategy is inherently time intensive, costly, and requires patient/family consent. A study by Lion et al evaluated communication experiences of patients and families preferring LOE in the emergency department by videorecording their visits.¹¹  This allowed for a robust evaluation of touchpoints as well as interpreter use.¹¹  Our communication tracking tool represents a novel attempt to measure inpatient touchpoints for both patients preferring LOE and patients preferring English, demonstrating inequities based on preferred language.

We specifically focused on resident physician-provided updates because residents typically serve as the “front-line” physician communicators with hospitalized patients and families at our institution. Our study demonstrated that residents have fewer interactions with those preferring LOE, and that when these interactions do occur, professional interpretation modalities are often underused. These findings are significant not just because of the obvious inequities that they reveal, but also because resident physicians are trainees who are actively refining communication skills that they will use for the remainder of their careers.^28–30 

Importantly, the touchpoints that occur for families throughout a patient’s hospitalization include not just interactions with physicians, but also with other staff, such as nursing and respiratory therapists. The inequities in resident touchpoint communication demonstrated in our work raise concerns that families preferring LOE may also experience such inequities with other providers. This matters because each of these touchpoints represents not just a clinical task, but also an opportunity for patients and families to learn, ask questions, and potentially intercept medical errors.²⁰  If families preferring LOE are experiencing inequities in touchpoints, then it is possible they are also being deprived of the opportunity to engage in these important “ad hoc” conversations with their team.

One of the core touchpoints in hospitalization is family-centered rounds, and our study found that patients and families preferring LOE are present less frequently during rounds compared with their counterparts preferring English, a finding congruent with previous studies.^18,31,32  Questionnaire data also indicated that 91% of residents reported sometimes, rarely, or never using interpretation for rounds. In light of these results, we found it interesting that the majority of residents (70%) reported rarely or never communicating with patients and families preferring LOE without interpretation. To explain this discrepancy, we posit that patients and families preferring LOE may not even be invited to participate in rounds, and thus rounds may not be seen by residents as a time when there was an interaction with families that would require the use of interpretation. A limitation of our study is that we are unable to delineate exactly why families were not present on rounds (i.e., whether it was because they were not invited, not present in the hospital, or another reason). Clearly, there is a need for ongoing work to support the involvement of families preferring LOE in family-centered rounds.

Questionnaire results also indicated that universally, residents felt that time limitations were a barrier to using interpretation with patients and families preferring LOE, despite noting comfort with both telephonic and in-person modalities. We suggest that residents are overall well-intentioned in their desire to communicate equitably with patients preferring other languages, but that their efforts are hampered by systemic factors, including time constraints. Developing creative solutions is crucial, given the reality of the fast-paced clinical environment. To minimize time required to coordinate interpretation, institutions should consider investing in video-remote interpretation modalities, ideally colocating these devices with patients in their rooms. This is particularly important given that recent studies have demonstrated that clinician use of interpretation increases with the availability of video-remote options.^11,33  Using inpatient navigators for patients preferring LOE may also be helpful because these individuals could be responsible for coordinating in-person interpretation and serve as a link for families to the clinical care team. Finally, we suggest creating and piloting clinical teamwork strategies, similar to the efforts that have been made to promote safe sign-out, such as protecting certain times for interpretation and creating interpretation utilization workflows.^34,35 

Future research efforts in this space should take a 2-pronged approach. First, there is a need for the development of reliable sources of data to describe disparities in communication interactions based on patient language preference. Although we selected to obtain data via self-report, other potential modalities include consistent documentation in the electronic medical record, observations, or videorecording of inpatient clinical encounters.¹¹  Rigorous assessment of the electronic medical record to ensure that patient language preference is accurately documented is also critical because previous research has demonstrated discrepancies.^36,37 

Second, quality improvement efforts should be used to (1) reduce inequity in communication based on patient preferred language and (2) improve use of interpretation. Several successful initiatives have been implemented to improve rates of interpretation, which could serve as models for other institutions.^7,38–40  These efforts have leveraged provider education and health care information technology interventions, such as electronic medical record alerts, to reduce clinician barriers to using interpreter services.^38–40  We also advocate for incorporating the voices and opinions of patients and families preferring LOE, as well as interpreters and trainees in developing interventions.^41,42 

Our study has several limitations. First, it was conducted at a single center, and thus results may not be generalizable. Second, we focused specifically on the resident perspective regarding interactions with patients and families preferring LOE and did not evaluate interactions by other health care providers, such as attending physicians and nurses. Our study also did not capture the impact of various patient and hospital factors, including length of stay, complexity, family health literacy, socioeconomic status, and hospital census, all of which may have an impact on communication patterns with families. Quality of interpretation is another important factor that our work does not measure. An additional limitation is the response rate to the survey, which was less than 50%, so there may be a sampling bias among the respondents. Finally, the communication tracking tool is reliant on self-reported data, and thus is subject to social desirability bias. However, although it is possible that residents did not accurately fill out the tool, if this occurred, we would expect that inaccuracies would lean in favor of more equitable communication. We argue that this makes our study’s results even more profound; if we assume that there was at least some social desirability bias present, it is possible that communication with families preferring LOE is even more inequitable than our results suggest.

In this study of resident communication with patients and families preferring LOE, we found that hospitalized patients and families preferring LOE are less likely to be present on rounds or to receive a resident update after rounds. When communication events do occur, residents report underuse of appropriate interpretation modalities.

Our results demonstrate inequity in communication for a patient population that continues to grow in the United States. Future efforts should focus on better understanding the root causes of this inequity and facilitating quality improvement initiatives, and fostering skills-based education for trainees. Key to this effort is developing partnerships with families preferring LOE and promoting adequate support for this population, both inside and outside of the hospital. Ultimately, creative, team-based, family-centered solutions are needed to ensure that residents are able to provide equitable communication with patients and families who prefer LOE.

The authors thank Alyssa Ciarlante and Feihan Xin for their assistance with statistical analyses for this study.